Releases: jippi/hashi-ui
v0.22.0
This release was sponsored by @seatgeek
Changes
- [nomad] evaluation list now supports filters
- [nomad] the job column now have a
healthcolumn, which is checking if the desired number of allocations are actually running for the job - [nomad] only show the deployment canary actions if the deployment have canaries
- [nomad] allow using Hashi-UI actions over HTTP (webservice style) - see
/API.md - [nomad] improve the allocation consul health check logic and UX
- [nomad] better use of the screen width for job view, now actions will show up as buttons if there is available space
- [shared] cleanup JS dependencies
- [shared] upgrade JS dependencies
v0.21.2
v0.21.1
v0.21.0
This release was sponsored by @seatgeek
Come work with me, we are hirering senior & junior system engineers and developers in NYC!
Fixes
- Missing Nomad / Consul connection check during startup caused panic (#331)
- Consul Read-Only did not work (#332)
- Hopefully the last fix for the proxy address rewriting (i've tested with Fabio and nginx)
Changes
- Hashi-UI now exposes a
/_statusendpoint for LBs and general health of Consul / Nomad - see README - Drained nodes will no longer be considered in cluster wide statistics (#333)
- You can now control the Consul and Nomad topbar color - see README (#334)
- All JS components are now lazy loaded, giving a much faster initial load
- All JS components are now loaded from an absolute path rather than relative, meaning you will no longer download the same files over and over when opening links in new tabs (#339)
- You can now restart a nomad job (e.g. deregister + register) (#344)
- Nomad Cluster page now has area charts for CPU and memory, showing up to 5min of ~live data rather than just the current values in a piechart (#338)
- Changes the color scheme for allocation statistics in Nomad Jobs overview
Special thanks to @hsmade that continue to be awesome in adding new features!
v0.20.1
v0.20.0
This release was sponsored by @seatgeek
Come work with me, we are hirering senior & junior system engineers and developers in NYC!
New contributor
I'm super happy to announce that @hsmade now has commit-bit ! :)
Feature: Consul Health integration ( @jippi )
If you have enabled the consul backend, then allocation lists will now show the Consul service health.
With the rewrite of the Go backend in 0.19.0, integrations like this is significantly easier to implement and maintain, expect most cross-nomad/consul integrations in the future.
Feature: Allocation resource statistics ( @hsmade )
You can now see ~live (1s resolution, for 5min) CPU and Memory statistics for an allocation.
In all allocations list, next to the "logs" shortcut in the actions column, there is a brand new "stats" icon taking you directly to this page. You can also navigate to it from a single allocation page -> stats in the menu.
Feature: Allocation port bindings ( @hsmade )
On the allocation info page, you can now see the IP and Port of your allocation.
Feature: Cluster & Client stats now show allocated resources ( @hsmade )
The client and cluster pie charts now also include how many resources you have locked to jobs
Cluster resource usage
Client resource usage
Feature: Client list shows CPU usage ( @jippi )
Every 3s the CPU usage of each node will be updated.
v0.19.0
This release was sponsored by @seatgeek
Come work with me, we are hirering senior & junior system engineers and developers in NYC!
Release notes
Breaking changes
- NewRelic has been removed from this release, as its value was very limited in debugging anyway
- Changed logging framework to
sirupsen/logrus
Rewritten Go backend
The Go backend has been completely rewritten, now Nomad and Consul share much much more code, making it way easier to do cross-system integrations, like showing Consul Health Checks in Nomad Allocation View and so on.
The rewrite has fixed a lot of minor bugs, like segfaults due to closed channels not being cleaned up, or missing checks for correct data.
As with any rewrite, there might be some new bugs in the rewrite, but the code in general is much simpler and easier to reason about - hopefully also making it easier for new contributors to understand the code and submit patches.
A major difference from previous releases is the way data is fetched from Consul and Nomad API.
Previous releases had 24/7 threads fetching and maintaining up-to-date data of all lists, even when no one was using the UI. This release no longer do this, but rather only fetch data when a user request the data, reducing the load on the Nomad and Consul endpoints significantly.
A consequence of the rewrite is also a more consistent error handling, where the Nomad and Consul errors will be relayed directly to the hashi-ui unmodified, rather than custom errors written by me. This also include errors like 404, which previously was ignored in error processing.
This change mean that people running Nomad 0.5.x trying to go to a Nomad Job single view, will see an error for 404 when hashi-ui tries to fetch Job Versions (which didn't exist until 0.6.x)
update: I've gotten reports from users that the memory requirements for hashi-ui has been reduced significantly with 0.19.0. The RAM consumption is stable around 50MB of RAM in a ~50 client cluster, where previous releases peaked at ~500MB of RAM in same setup.
Better visibility into resource subscriptions
Additionally, every connection will now log to console what resources it's subscribed to every 10s
Minor UI tweaks
- All overview pages now default put focus on the first input filter field (mostly the Job column)
- The clients list now show CPU utlliization, it's refreshed every ~10s
Docker
use jippi/hashi-ui:v0.19.0 to target this release